Species diversity is thought to stabilize functioning of plant communities. An alternative view is that stability depends more on dynamics of dominant species than on diversity. We compared inter-annual variability (inverse of stability) of aboveground biomass in paired restored and remnant tallgrass prairies at two locations in central Texas, USA. Data from these two locations were used to test the hypothesis that greater richness and evenness in remnant than restored prairies would reduce variability in aboveground biomass in response to natural variation in rainfall. Restored prairies were chosen to be similar to paired remnant prairies in characteristics other than species diversity that affect temporal variability in biomass. Variability was measured as the coefficient of variation among years (square root of variance/mean; CV), where variance in community biomass equals the sum of variances of individual plant species plus the summed covariances between species pairs. Species diversity over five years was greater by a factor of 2 or more in remnant than restored prairies because richness and evenness were greater in remnant than restored prairies. Still, the CV of community biomass during spring and CV of annual biomass production did not differ consistently between prairie types. Neither the sum of species covariances nor total community biomass differed between prairies. Biomass varied relatively little in restored compared to remnant prairies because biomass of the dominant species in restored prairies (the grass Schizachyrium scoparium) varied less than did biomass of other dominant and sub-dominant species. In these grasslands, biomass response to natural variation in precipitation depended as much on characteristics of a dominant grass as on differences in diversity.